Related Experiment Videos
Truncated RAP-MUSIC (TRAP-MUSIC) for MEG and EEG source localization
Niko Mäkelä1, Matti Stenroos2, Jukka Sarvas2
1Department of Neuroscience and Biomedical Engineering (NBE), Aalto University School of Science, Espoo, Finland; BioMag Laboratory, HUS Medical Imaging Center, Helsinki University Hospital (HUH), Helsinki, Finland.
Neuroimage
|November 13, 2017
Summary
A new method, truncated recursively-applied-and-projected MUSIC (TRAP-MUSIC), accurately estimates the number of brain-signal sources. This TRAP-MUSIC approach enhances the reliability of MUltiple SIgnal Classification (MUSIC) for brain imaging.
Area of Science:
- Neuroscience
- Biophysics
- Signal Processing
Background:
- MUltiple SIgnal Classification (MUSIC) is used to locate electrically active brain regions using magnetoencephalography (MEG) and electroencephalography (EEG) data.
- Conventional Recursively Applied and Projected MUSIC (RAP-MUSIC) algorithms can inaccurately estimate the number of underlying brain signal sources.
Purpose of the Study:
- Introduce a novel MUSIC-based method, TRAP-MUSIC, to improve the accuracy of source number estimation.
- Enhance the robustness and reliability of brain source localization using MEG and EEG data.
Main Methods:
- Developed truncated recursively-applied-and-projected MUSIC (TRAP-MUSIC) by incorporating sequential dimension reduction into signal-subspace projection.
- Compared TRAP-MUSIC with conventional RAP-MUSIC through simulations with varying signal-to-noise ratios, source time-course correlations, and initial dimension estimates.
- Validated TRAP-MUSIC performance using real-world MEG data.
Main Results:
- TRAP-MUSIC significantly improves the performance of MUSIC-type localization compared to RAP-MUSIC.
- The method successfully and robustly identifies active brain regions and accurately estimates their quantity.
- TRAP-MUSIC demonstrates improved performance across various simulation parameters.
Conclusions:
- TRAP-MUSIC offers a more reliable approach for estimating the number of brain signal sources.
- This method enhances the accuracy and robustness of brain source localization.
- TRAP-MUSIC is suitable for both online and offline applications in MEG and EEG neuroimaging.